An aerospace part without NADCAP traceability is undeliverable product — iLEAN assembles the file, not your quality team.

The problem

The part's file lives in six places — and nobody stitches it together until the auditor calls.

The aerospace quality manager knows the routine by heart: the work order is in the ERP, the material trace in a warehouse spreadsheet, the real furnace parameters in the heat treat subcontractor's SCADA, the penetrant results in the PDF the NDT lab emailed over, the plating supplier's CofC in a drawer, and the operator's signature on the router that travels with the part. Stitching all that together for a single part takes hours. For a batch of 50, it takes a week. For a NADCAP audit that asks for 30 parts at random, it takes the whole month before.

The NADCAP file is not done badly — it is done late. And when it is done late, it is done with disproportionate effort and with the fear that something is missing. The data existed at the time, but it was not captured at the time; you try to rebuild it, and in aerospace rebuilding is exactly what the auditor does not want to see.

How it fits the IRIS system

iLEAN does not add a seventh system — it seals the cracks between the six you already have.

The problem with NADCAP traceability is not missing information: it is information that lives on islands and that, at the moment of truth (the part going out for delivery, the auditor asking for a random dossier), does not reach the person who signs it already stitched together. iLEAN acts as the putty that fills those cracks, without asking you to change your ERP, your MES, the furnace SCADA or the subcontractor's procedure.

Tracer stitches traceability part by part. Writer assembles the CofC in the customer's format. Vision reads furnace curves and NDT records as evidence. The person signs — never the other way round.

The three iLEAN pieces applied to NADCAP traceability:

• iLEAN Tracer — an agent that keeps the file of every part alive from raw material intake (heat lot, mill cert) through to delivery. It cross-references the work order, the router, the operations performed and the operator who signed each one, and links the evidence arriving from outside (subcontractors, NDT labs). There is no need to "close out" the file — it is always closed and always alive.
• iLEAN Writer — an agent that assembles the CofC in the format each customer asks for (OEM CofC, FAA 8130-3 where applicable, EASA Form 1, the tier-1's internal format). The quality team does not open templates: they review the document already assembled, check that the evidence adds up, and sign. If something is missing, the batch is not marked deliverable.
• iLEAN Vision — a vision terminal over the special-process points: it reads the heat treat furnace curve at the end of the cycle, reads the photographic record of the penetrant inspection, reads the data matrix marked on the part to link the operation to the work order. It works with no network: if the plant loses connectivity, it keeps capturing and links up when the connection returns. The visual evidence is stored cryptographically signed in the file.

See the full IRIS architecture →

Before and after

A rebuilt file vs. a live file with iLEAN

Aspect	Manual traceability + spreadsheets + email	With iLEAN Tracer + Writer + Vision
Heat lot ↔ part	Handwritten note on the router	Automatic link by data matrix, read by Vision
Real heat treat furnace parameters	Subcontractor's PDF arriving by email	Curve read by Vision or captured by Connect, attached to the file
Subcontractor's CofC	PDF in a drawer / shared folder	Captured by Connect at second zero, linked to the part
CofC to the customer	Word template the technician fills in	Assembled by Writer in the customer's format; a person reviews and signs
NADCAP audit: dossier for 30 random parts	The whole month before spent rebuilding	Minutes per part; evidence already stitched and signed
Part with an incomplete file	Discovered at delivery or during the audit	Batch held before it can be marked deliverable

Impact estimate

Impact estimate for your plant — to be validated with your numbers.

The block below is an estimate to be validated with the specific data of your plant. We put it forward so the committee has an order of magnitude; we refine it during the diagnostic.

• Tier-1 plant machining structural parts, NADCAP heat treat subcontracted, penetrant NDT in house or external, plating subcontracted. Volume in the order of hundreds of parts/month with several references running in parallel.
• Tracer + Writer + Vision pilot on one specific part family (the customer Pareto — usually a recurring part for a large OEM). First value expected within a few weeks: a live dossier for that family instead of ad-hoc reconstruction.
• Indicative payback between 4 and 9 months. Levers: quality time rescued from stitching dossiers, less risk of undeliverable product, NADCAP audit preparation in hours instead of weeks, less documentary rework from paperwork non-conformities.
• A ≥50% reduction in audit preparation time is a prudent target; the plants that have already seen it in operation bring it down further.

And the aerospace CAIO's reasonable doubt

"What if the AI makes up a data point on the CofC?" — hallucination is a problem of free generation, not of anchored tasks. Writer does not "write" the CofC: it assembles it from evidence already captured (material heat lot, furnace curve, NDT record, operator signature). In tasks anchored to the source, the best models brought error below 1.5% ^[1]. And even so, what is critical is never decided alone: the CofC goes out with a human signature, and iLEAN's three safety rings guarantee that no agent touches the plant's critical work order — documentation goes out cryptographically signed from ring 2 outward, not the other way round.

^[1] OpenAI paper "Why Language Models Hallucinate", 2025 — on the reliability of AI in anchored tasks.